Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
ISSN: 2695-2246 STRIDES EDUCATIONAL FOUNDATION
This paper and more can be downloaded without charge from http://www.bequarterly.rysearch.com
Are Latifat Olasubomi: [email protected] a. Research Officer, Small Business & Informal Economic Research Group, Department of
Economics, University of Lagos Akoka, Akoka Lagos
How to cite this article: Are, L. O., (2019). Currency Devaluation and Trade Balance Nexus: A Test
of Marshall-Lerner Condition in Nigeria. BizEcons Quarterly, 4, 23–43.
Journ
al h
om
e pag
e: h
ttp:/
/ww
w.b
equar
terl
y.r
yse
arch
.co
m
Journ
al h
om
e pag
e: h
ttp:/
/ww
w.b
equar
terl
y.r
yse
arch
.co
m
Are Latifat Olasubomia
Currency Devaluation and Trade Balance Nexus: A Test of
Marshall-Lerner Condition in Nigeria
A b s t r a c t
Devaluation on the trade balance in recent times emerged as
a vital policy issue in Nigeria due to the recent economic
crisis. This paper examines the nexus between devaluation
and Nigeria’s trade balance by empirically investigating the
Marshall-Lerner condition using annual data covering the
period of 1986-2015. It also examines the dynamics of the
relation between Nigeria’s trade balance and the real
exchange rate. The Engel-Granger OLS-based
Cointegration technique and the Error Correction
Mechanism (ECM) modelling technique were employed to
conduct these analyses. Findings suggest that the Marshall-
Lerner condition is not satisfied in Nigeria and that real
exchange rate has no significant effect on the trade balance
both in the long-run and short-run. Also, findings revealed
that income effects dominate price effects on trade in
Nigeria.
Received: December 28, 2018
Revised:
February 3, 2019
Accepted:
June 1, 2019
Keywords:
Exchange rate
Currency devaluation
Trade balance
JEL Classification:
B41
C21
F31
1. Introduction
In many countries, including Nigeria, the deliberate downward adjustment of the
official exchange rate, has been used to address various macroeconomic issues such
as increasing international competitiveness, correcting a currency overvaluation,
reducing a current account deficit, and in most cases, improving the trade balance
(see Loto, 2011: 624-633). It is also stipulated in the International Monetary Fund
(IMF) policy that a country should devalue its currency to solve any essential
disequilibrium issues in its balance of payments.
Volume 4 (2019) • 23 – 43
Are B I Z E C O N S Q U A R T E R L Y
• 24 •
From 1960 to early 1970s, Nigeria was predominantly an agrarian economy
depending on the agricultural sector. During this period, a fixed exchange rate regime
was practised with the Naira fixed at par to the British Pounds Sterling. After the
discovery of Oil in Nigeria, followed by the oil boom period in the early 1970s, the
oil sector relegated the agricultural and other sectors of the economy to the ground to
become the backbone of the economy accounting for 90% of export, 80% of
government revenue and 85% of foreign exchange earnings. However, when the oil
price crashed, the economy was faced with serious internal and external imbalances
that led to macroeconomic instability. In response to this macroeconomic instability,
the Federal Government adopted the Structural Adjustment Programme (SAP)
proposed by the IMF in 1986 to tackle these crises.1 Following the adoption of the
SAP, the Naira was devalued to achieve internal and external balance in the medium
term, which marked the beginning of the flexible exchange rate regime in Nigeria.
After the implementation of SAP, the economy did achieve both a favourable balance
of payments position and an improved macroeconomic condition.
In 2014, history repeated itself and the international oil price crashed from as high as
$115 per barrel in June 2014 to as low as $35 at the end of February 2016, which
plunged the economy into recession (Evans, 2019). Because proceeds from oil export
revenue still account for more than 70% of government revenues and about 90% of
foreign exchange earnings, this crash engendered massive distortions in the economy.
However, even though this current crisis is similar to what was experienced in the
early 1980s, the Central Bank of Nigeria (CBN) resisted against devaluing the Naira
as advised by some economists and the IMF. The argument put forward by the CBN
governor and others who were against the devaluation of the Naira was that the
structure of the present Nigeria economy and Foreign trade structure did not support
devaluation improving the trade balance and creating external balance.
In Nigeria foreign trade structure is characterized by major export of crude oil,
however, crude oil prices do not respond to policy instruments especially in the short
run because they are inelastic (Odili, 2007: 21). This controversial debate on the
economic notion of using a devaluation to achieve economic growth by improving
the trade balance, before now, has attracted the attention of many researchers who
have addressed this contentious issue by evaluating the impact of devaluation on the
trade balance either by directly examining the impact of devaluation on trade balance
or by investigating the Marshall-Lerner condition. There are various empirical studies
that have attempted to examine the validity of Marshall-Lerner condition in Nigeria
with no consensus as to what the impact of devaluing the Naira will be on the trade
balance. Hence, the need to re-evaluate how and what is the impact of devaluing the
Naira on Nigeria’s trade balance.
1 The Structural Adjustment Programme is as economic reform package aimed at enhancing
local productivity, achieving balance of payments equilibrium, expanding economic base and
boosting the growth potential of developing countries.
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 25 •
Therefore, this paper seeks to contribute to the existing literature by critically re-
examining the impact of devaluing the Naira on Nigeria’s trade balance. This was
done by empirically investigating the M-L condition using Nigeria’s total imports
and exports data and also by empirically analysing the long-run and short-run
relationship between the trade balance and the real exchange rate.2
The remaining sections of this paper are organised as follows: section II provides a
review of relevant existing literature; section III explains the data and method;
section IV discusses the empirical results while section V provides the conclusion
and policy implications.
2. Literature Review
2.1 Theoretical Framework
The question of how devaluation impacts trade (current account) balance or the
balance of payments in general has been addressed over the decade using these three
different approaches; the elasticity approach to trade balance, the absorption
approach to trade balance, and the monetary approach to balance of payments. This
paper employed the conventional elasticity approach to trade balance.
The traditional elasticity approach is the most commonly adopted approach by many
empirical studies in assessing the impact of a devaluation on the trade balance. The
elasticity approach focuses on the degree of responsiveness of import and export
volumes to a change in relative prices through the exchange rate. This approach was
propounded by Bickerdike (1920), and popularised by Robinson (1937) and Metzler
(1948).
According to the elasticity approach, devaluation has two effects on the trade balance
which are the value effect and volume effect, and the net effect depends on which of
the two effects dominate. Following a devaluation, the domestic price of imports and
exports are affected as follows: The foreign price of domestic export falls relative to
foreign prices, while the foreign price of imports rises relative to domestic prices,
both in proportion to the devaluation. This makes the domestic exports cheaper for
foreigners and foreign imports more expensive for domestic residents. The volume
effect occurs when foreigners demand more of the domestic export which has
become relatively cheaper while domestic residents reduce their demand for foreign
import and substitute that with cheaper domestic good.
However, while less is demanded of foreign goods, more money is spent to purchase
the same quantity or less due to the fall in the purchasing power of the domestic
currency relative to the foreign currency. This translates into the value effect that
only worsens the current account. Therefore, if the value effect dominates the volume
effect after a devaluation, which is mostly the case in the short-run, the trade balance
2 Exports in Nigeria are usually segregated into oil and non-oil with oil exports accounting for
more than 90% of the total exports. Also, the trade balance can be divided into trade balance
with oil and trade balance without oil.
Are B I Z E C O N S Q U A R T E R L Y
• 26 •
will worsen. On the other hand, should the volume effect dominate, which is more
likely to occur in the long-run due to an adjustment lag that takes place in the short-
run following a devaluation, the trade balance will improve.
According to Alexander (1952: 263), the effect that dominates both on the export and
import side depends on these four elasticities; (1) elasticity of foreign demand for the
devaluing country’s export, (2) elasticity of domestic demand for foreign import, (3)
elasticity of domestic supply of export and (4) elasticity of foreign supply of import.
Based on these four elasticities, the Bickerdike-Robinson-Metzler (BRM) model
provided a sufficient condition (BRM condition) for a trade balance improvement
following a devaluation. From the BRM condition, Marshall (1923) and Lerner
(1944) described the condition under which a devaluation will promote the trade
balance based on the following assumptions: that the price elasticity of supply at
home and abroad both equal infinity, ignoring the monetary effects of exchange rate
variations, assuming an initial balance of trade and that import, and export depends
on exchange rate variations (see Loto, 2011: 624-633).
Hence, the M-L condition stipulates that, devaluation will improve the trade balance
if the export and import volumes are sufficiently elastic with respect to relative
prices. The condition requires that the sum of the trade elasticities to changes in
relative prices be greater than unity in absolute terms, and the satisfaction of this M-L
condition guarantees devaluation will improve the trade balance. The M-L condition
can be mathematically expressed as𝜂𝑥 + 𝜂𝑚 > 1, where 𝜂𝑚and 𝜂𝑥are the price
elasticity of demand for imports and exports respectively.3
2.2 Empirical Literature Review
In international economics, the traditional elasticity approach to devaluation has been
the most widely adopted theoretical approach to evaluating the impact of devaluation
on the trade balance. This approach forms the basis for the M-L condition which is
what most studies empirically investigate for different countries. The M-L condition
requires that the sum, in absolute terms, of the import and export price elasticities be
greater than one for the trade balance to be improved by devaluation.
Robinson (1937) described the condition under which the trade balance will be
improved due to a change in the exchange rate and also derived the corresponding
elasticities which have formed the foundation for various empirical analysis ever
since (see Eita, 2013: 511-518). Several studies have employed different techniques
in estimating these elasticities to test if M-L condition holds for different countries
with mixed results.
Goldstein and Khan (1978) used the full-information maximum likelihood method to
estimate reduced form export demand and export supply functions simultaneously for
eight industrial countries (viz., Belgium, France, Germany, Italy, Japan, the
3A step by step derivation of the mathematical expression of the M-L condition is provided in
Appendix C.
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 27 •
Netherlands, the United Kingdom, and the United states) using quarterly data over
the period of 1955-1970. The price elasticities estimated from these functions were
combined with the elasticities of demand for import and export estimated in
Goldstein and Khan (1976) for the same countries. The combined results provided
support for the M-L condition for all countries except for Japan. However, a study by
Sinha (2001) estimated the price and income elasticities for five Asian countries
(Japan, Philippines, Sri Lanka, India and Thailand) using different techniques and
found that the M-L condition is met for four of those countries including Japan.
Miles (1979) established a direct link between the trade balance and the exchange
rate and employed several econometric techniques to examine the effect of
devaluation on the trade balance for 16 devaluations in 14 countries, both DCs and
LDCs, over the period of 1967-1962. Miles concluded that devaluation has no
significant improvement effect on the trade. In contrast, Himarios (1989), after
criticising Miles's model of using the trade balance to income ratio has the dependent
variable rather than the trade balance, employed the OLS method to analyse 60
devaluation experience for 27 countries from two different samples: 1953–73 and
1975–84. The author found that changes in the relative price as a result of
devaluation, has significantly positive effect on the trade balance.
Taking a step further from what was done by these two studies, Bahmani-Oskooee
(1994) employed the cointegration and error correction modelling technique to re-
examine the long-run and short-run relation between the trade balance and the real
effective exchange rate of 19 DCs and 22 LDCs using quarterly data from 1971 to
1990. The result from this study showed that the trade balance and the exchange rate
have no significant long-run relationship for most of the countries which is consistent
with the result of Miles (1979).
Arinze (1994) employed the cointegration technique to test if a relation exists
between the trade balance and the exchange rate in the long run for nine Asian
countries using quarterly data from 1973Q1 to 1991Q1. Contrary to the results above,
this study found a significant long-run equilibrium between the trade balance and the
real exchange rate. Likewise, Bahmani-Oskooee (1998), employed a Maximum-
likelihood cointegration technique to estimate the trade elasticities for six LDCs
using quarterly data over the period of 1973I-1990IV. From those estimates, it was
concluded that the trade balance of these counties did not improve by devaluing their
currencies.
The literature review for Nigerian studies indicate lack of consensus on how
devaluation will impact the trade balance and if the M-L condition does or will holds
in Nigeria. For example, Odili (2014) employed an autoregressive distributed lag
(ARDL) cointegration estimation technique to analyse the long-run and short-run
relationship that exists between the balance of payments and some macroeconomic
variables from 1971-2012. While using the export and import estimates to test if the
M-L condition holds for Nigeria, the study revealed that a positive and significant
Are B I Z E C O N S Q U A R T E R L Y
• 28 •
relationship exists between the exchange rate and the trade balance and found support
for the M-L condition in Nigeria.
In corroboration with this finding is the result of Ogbonna (2011) which found that
M-L condition hold for Nigeria and devaluation can be used to improve the trade
balance. Ogbonna (2011) employed the Johansen cointegration approach to estimate
a multivariate cointegration system to examine the empirical relationship between the
real exchange rate and the trade balance both in the long-run and short-run, over the
period of 1970-2005. Results from this study confirm that the M-L condition holds
for Nigeria.
Similarly, Igue and Ogunleye (2014), using the OLS method, Johansen cointegration
technique and vector error correction methodology, estimated an import demand
equation, export supply equation and a trade balance equation using quarterly data
series from 1985Q1 to 2010Q4. The results from the study confirmed that the M-L
condition is satisfied for Nigeria and also established that if the exchange rate is
devalued by just 1%, the trade balance will improve by at least 1.16% implying that
devaluation will significantly improve the trade balance in the long-run.
In contrast, Ogundipe, Ojeaga and Ogundipe (2013) empirically investigated the
impact of devaluing the Naira on the trade balance over the period of 1970-2010
using a Johansen cointegration and variance decomposition analyses. They found that
exchange rate induces an inelastic and significant effect on the trade balance and
therefore worsens the trade balance in the long-run. In support, Loto (2011)
investigated the M-L condition in terms of total imports and non-oil exports during
the period of 1986-2010, using the OLS method. The results confirmed that
devaluation cannot improve the trade balance (without oil) due to the structure of the
Nigerian economy and its foreign trade.
Also, Eke, Eke and Obafemi (2015), using larger time period covering 1970 to 2012,
spanning both the pre and post 1986 devaluation of the Naira, used the Johansen
cointegration technique and error correction mechanism modelling technique to
estimate the effect of on the trade balance of exchange rate. The estimated results
from this study suggested that devaluation cannot lead to an improvement of the trade
balance. Thus, providing support for other negative results.
Similarly, Anoke, Odo and Ogbonna (2016), examined the effect of a depreciation on
the trade balance using the cointegration test, Vector Error Correction Model and
other technique from 1986-2014. The authors concluded from their estimates
Nigeria’s trade balance cannot be improved in the short run from depreciation of the
Naira, as only countries that are initially export based before deprecation only benefit
from such depreciation which is similar to the point raised by Loto (2011), as an
argument of lack of support for M-L condition in Nigeria.
However, this ongoing debate on the relationship that exists between the Nigeria’s
trade balance and a devaluation of the Naira calls for a re-evaluation of this
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 29 •
relationship and the M-L condition given the current situation of the Nigerian
economy.
3. Data and Methods
3.1 Models
The elasticity approach to the trade balance will be employed in preference to the
other approaches, because the focus is to empirically examine the M-L condition and
it is the conventional approach generally adopted in the literature to investigate the
M-L condition (see, Ogbonna, 2011; Igue et al. 2014). Hence, employing this
approach provides a basis for situating our findings within the existing literature.
To estimate the necessary elasticities, both an import demand and export supply
model were specified. In addition, a trade balance model was also formulated and
estimated as specified in existing literature (e.g. see Igue et al, 2014: 347-358)
The import demand model is expressed as a function of domestic income and the
relative price of imports, while the export supply model is expressed as a function of
world income and relative export prices as shown below in their most general forms:
The import demand model can be specified in a functional form as;
𝑀𝑡 = 𝑓(𝑌𝑡 , 𝑅𝑃𝑀𝑡) (1)
𝑋𝑡 = 𝑓(𝑌𝑊𝑡 , 𝑅𝑃𝑋𝑡) (2)
To have the estimated coefficients interpreted in elasticity form, the models are
expressed in a log-linear form as long-run import and export equations:
𝑙𝑛𝑀𝑡 = 𝛽0 + 𝛽1𝑙𝑛𝑌𝑡 + 𝜂𝑚
𝑙𝑛𝑅𝑃𝑀𝑡 + 휀𝑡 (3)
𝑙𝑛𝑋𝑡 = 𝛿0 + 𝛿1𝑙𝑛𝑌𝑊𝑡 + 𝜂𝑥
𝑙𝑛𝑅𝑃𝑋𝑡 + 𝜇𝑡 (4)
where 𝑀𝑡and 𝑋𝑡 are the volume of domestic imports and exports during year 𝑡 to the
world respectively; 𝑌𝑡 and 𝑌𝑊 are the domestic and world income in period 𝑡
respectively; 𝑅𝑃𝑀𝑡and 𝑅𝑃𝑋𝑡are the relative price of imports and exports in period
𝑡 respectively; 휀𝑡and 𝜇𝑡are the error terms for which standard assumptions are made;
𝛽1& 𝛿1 represent the elasticities with respect to income, while 𝜂𝑚
& 𝜂𝑥
, represent the
elasticities with respect to relative prices.
These equations will be used to estimate demand elasticities for both imports and
exports with respect to relative prices and income on an aggregate and bilateral basis.
It is expected that:
𝜂𝑚 < 0, since a devaluation should cause the prices of imports to rise relative to
domestic prices, leading to an increase in relative price of imports.
𝜂𝑥 < 0 because devaluation should make domestic export prices fall relative to the
world’s export prices.
Are B I Z E C O N S Q U A R T E R L Y
• 30 •
𝛿1 > 0 such that when the world’ income increase, it is expected to increase the
demand for Nigerian exports.
𝛽1 >< 0, as the expected sign is ambiguous. The sign is generally taken to be
positive because as domestic income increases, demand increase and the volume of
import demand increases. However, it could be negative if the rise in real income is
due to increase in national output or production, especially when these outputs are
substitutes for imported goods. Then, imports will decrease as the domestic income
increases.
In other for the M-L condition to be satisfied it is expected that the absolute value of
the sum of the import and export demand elasticities be greater than one. Hence, we
test the following hypothesis;
Null hypothesis = 𝐻0: |𝜂𝑚
+ 𝜂𝑥
| = 1
Alternative hypothesis = 𝐻1: |𝜂𝑚
+ 𝜂𝑥
| > 1
Another model formulated and estimated in this paper is a trade balance model which
does not formally test for the M-L condition. But it used to determine the effect of the
exchange rate on the trade balance.
However, there has been debate as to the appropriate measure of the trade balance.
Several studies e.g. (see Bahmani-Oskooee, 1991:403-407) measured the trade
balance as the ratio of export to imports and has convincingly argued that it is a more
convenient and appropriate measure as it is insensitive to the units of measurements
of exports and imports and also insensitive to whether the imports and exports value
are in domestic or foreign currency. This method of measuring the trade balance is
also employed in this study and expressed thus in line with Boyd and Smith (2001:5):
𝑇𝐵𝑡 = (𝑋𝑡𝑃𝑡)
(𝑀𝑡𝐸𝑡𝑃𝑡∗)⁄ (5)
where, 𝑇𝐵𝑡 is the trade balance measured as the ratio of the export volume
𝑋𝑡 multiplied by the domestic prices 𝑃𝑡, to import volume 𝑀𝑡 multiplied by the
foreign prices 𝑃𝑡∗ and the nominal exchange rate 𝐸𝑡.
The trade balance equation to be analysed in line with the M-L condition prescription
is specified thus:4
𝑙𝑛𝑇𝐵𝑡 = 𝜃0 + 𝜃1𝑙𝑛𝑅𝐸𝑅𝑡 + 𝜃2𝑙𝑛𝑌𝑡 + 𝜃3𝑙𝑛𝑌𝑊𝑡 + 𝜖𝑡 (6)
Where 𝜃 = (𝛽0 − 𝛿0); 𝜃1 = (𝜂𝑚 + 𝜂𝑥 − 1); 𝜖𝑡 = (휀𝑡 − 𝜇𝑡), 𝑌𝑡 is the domestic
income in period 𝑡; 𝑌𝑊𝑡 is the world income in period 𝑡; 𝑅𝐸𝑅𝑡 is a trade-weighted
real effective exchange rate, defined as the foreign price of the Naira in period t. For
the trade balance to improve from devaluation, it is expected that;
4The mathematical derivation of the trade balance equation estimated is provided in appendix
C.
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 31 •
𝜃1 < 0 , is expected to be significantly negative since a decrease in 𝑅𝐸𝑅𝑡𝑡 reflects a
devaluation of the Naira.
𝜃2 >< 0, because an increase in domestic income should increase import demand if
the increase in income is not as a result of an increase in the production of import
substitutes goods which will, in turn, worsen the trade balance. However, if increase
in the production of import substitutes brought about the increase in income, this will
improve the trade balance by reducing the volume of import demands.
𝜃3 > 0, since an increase in world’s income is expected to increase the demand for
Nigeria export and improve the trade balance.
3.2 Data
Data employed in this study were collected over the period covering 1986-2014
inclusive. The choice of 1986 was influenced by the fact that it marked the period
when the Naira was first devalued following the adoption of the IMF’s Structural
Adjustment Programme (SAP). This ensures our empirical analysis is over a period
governed by a flexible exchange rate regime in Nigeria. The data collected, and their
sources include: Nigeria’s total imports and exports volume index sourced from
UNCTAD STAT; Nigeria’s real GDP in constant local currency units; the world’s
real GDP in constant 2005 US$; the GDP deflator for Nigeria drawn from IMF
international Financial Statistics; the world export unit index sourced from IMF
international Financial Statistics; and the real effective exchange rate for Nigeria
collected from the World Bank database. Some of the variables constructed from
these data are defined as:
𝑙𝑛𝑅𝑃𝑀𝑡 which is the natural logarithm of the relative price of imports and it is
defined as the ratio of Nigeria import prices to the Nigeria GDP price deflator.
𝑙𝑛𝑅𝑃𝑋𝑡 which is the natural logarithm of the relative price of exports defined as the
ratio of Nigeria’s export price to World export unit index.
𝑙𝑛𝑅𝐸𝐸𝑅𝑡 which represents the natural logarithm of the real effective exchange rate of
Nigeria is constructed by dividing the nominal effective exchange rate (a measure of
the value of a currency against a weighted average of several foreign currencies) by a
price deflator or index of costs (World Bank database).
3.3 Methodology
Annual time series data are used in this study to empirically analyse equations (3), (4)
and (6) specified above. These time series data are examined to avoid encountering
common problems that time series data generally possess (Evans et al, 2018). The
most crucial of those problems for this analysis is the problem of non-stationarity.
Time series are said to be nonstationary because their probability distribution are not
stable overtime and as such exhibit a stochastic trend. A non-stationary time series
data is practically invalid for forecasting purpose. Also, regressing two or more
nonstationary time series may lead to spurious regression.
Are B I Z E C O N S Q U A R T E R L Y
• 32 •
Unit roots tests were conducted to confirm the stationarity of the individual data
series and also to determine the order of integration for each variable. The Dickey-
Fuller unit root test was generally employed, and the Augmented Dickey-Fuller test
was used in few cases where the error term in the equations are serially correlated.
The presence of unit-root in each variable are tested for using the following equation:
(1 − 𝐿)𝑙𝑛𝑍𝑡 = 𝜌0 + 𝜌1𝑙𝑛𝑍𝑡−1 + 𝜇𝑡 (7)
where, (1 − 𝐿)𝑙𝑛𝑍𝑡 is the first difference of the log of the time series; 𝑙𝑛𝑍𝑡−1is a one
period lag of the time series; 𝑡 is a trend variable; 𝜌0 is a constant term and 𝜇𝑡 is a
stochastic error term. Equation (7) depicts a random walk with drift while and the
decision of whether a time series is stationary or not depends on the estimates of 𝜌1.
A time series free of unit root will have a significant estimate of 𝜌1 < 0, whereas, a
time series with unit root or stochastic process will have an estimate of 𝜌1 that is
statistically not different from zero. Therefore, the following hypotheses are put to
test using the appropriate critical values:
Ho: 𝜌1 = 0
Ha: 𝜌1 < 0
These tests were conducted using the formula for a t-test. However, the distribution
of these tests is non-standard and special critical values devised by Dickey-Fuller
must be used for inference purposes.
After identifying non-stationary data and determining their level of stationarity of the
data, cointegration test is conducted to determine if a long-run relationship exists
between the set of nonstationary variables. A residual based cointegration test based
on the Dickey-fuller test is employed to determine if a linear combination of these
nonstationary series cancels out the stochastic trends in the series. The residual based
Dickey-Fuller test for cointegration is employed to test for cointegration between the
variables in the import demand, export supply and trade balance equations. This is
done by regressing each of the three equations using OLS method and subjecting the
residuals from each set of regression to a unit root test to determine if the residual is
integrated of a lower order of integration than the order of integration of the variables
in the long-run levels model. The OLS regression technique is applied to the
following equation to estimate the residuals;
𝑙𝑛𝑀𝑡 = 𝛽0 + 𝛽1𝑙𝑛𝑌𝑖𝑡 + 𝛽2𝑙𝑛𝑅𝑃𝑀𝑖𝑡 + 𝜋𝑡 (8)
𝑙𝑛𝑋𝑡 = 𝛿0 + 𝛿1𝑙𝑛𝑌𝑊𝑖𝑡 + 𝛿2𝑙𝑛𝑅𝑃𝑋𝑖𝑡 + 𝜌𝑡 (9)
𝑙𝑛𝑇𝐵𝑡 = 𝜃0 + 𝜃1𝑙𝑛𝑅𝐸𝑅𝑡 + 𝜃2𝑙𝑛𝑌𝑡 + 𝜃3𝑙𝑛𝑌𝑊𝑡 + 𝜎𝑡 (10)
where, all variables remain as earlier defined; 𝜋𝑡, 𝜌𝑡 , 𝜎𝑡 are the estimated residuals
from each regression which are tested to confirm the presence of cointegration in
each model.
The residuals from these regressions represent a linear combination of these variables
at levels and it is this linear combination that should be integrated of order zero to
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 33 •
ensure a cointegrated relationship. If it is the case that the residuals are integrated of
order zero, the regression is not spurious because it implies that the variables are
cointegrated and there exists a long-run equilibrium relationship amongst the
dependent and the independent variables in each regression.
The estimated residuals from each regression are then subjected to unit root test. The
Dickey-Fuller test in the context of cointegration is known as the Engle-Granger or
Augmented Engle-Granger with no intercept and no trend term because the estimated
residuals are normally distributed with zero mean and constant variance i.e. NIID
(0,1). Hence the estimated residuals are tested for unit root using the model below:
(1 − 𝐿)𝜖𝑡 = 𝜔1𝜖𝑡−1 + 𝜇𝑡 (11)
where, 𝐿 is the lag operator; 𝜖𝑡−1 is a one period lag of the estimated residual; and 𝜇𝑡
is a stochastic error term. The decision of whether a cointegration exists between the
regressor and the regressand in each model depends on the estimates of 𝜔1. Hence,
we test the following hypothesis using the appropriate critical value:
Ho: : 𝜔1 = 0
Ha: : 𝜔1 < 0
The Mackinnon critical values for cointegration used in testing this hypothesis are
separately computed at 1%, 5% and 10% level of significance using this response
surface regression:5
𝐶𝑘(𝑝) = 𝛽∞ + 𝛽1𝑇𝑘−1 + 𝛽2𝑇𝑘
−2 + 𝑒𝑘 (12)
where 𝐶𝑘(𝑝) denote the estimated 𝑝% quantile from the 𝑘𝑡ℎ experiment, 𝑇𝑘denotes
the sample size, 𝛽∞is an estimate of the asymptotic critical value for a test at level 𝑝,
𝛽1and 𝛽2determine the shape of the response surface for finite values of 𝑇. 6
Using the Mackinnon critical values computed, the variables will be cointegrated if
the null hypothesis is rejected in favour of the alternative which implies that the
estimated residual is unit root.
After confirming if the variables in each of the three models are cointegrated, we
estimate an error correction mechanism model for the import demand, export supply
and trade balance equations. The error correction mechanism model is usually
formulated as follows;
(1 − 𝐿)𝑌𝑡 = 𝜑0 + 𝜑2(1 − 𝐿)𝑙𝑛𝑍𝑡 + 𝜑0𝐸𝐶𝑀𝑡−1 + 𝑣𝑡 (13)
where, (1 − 𝐿) is the first difference operator; 𝐸𝐶𝑀 is the estimated residual from the
cointegration regression. 𝑌𝑡, is the dependent variable which includes 𝑙𝑛𝑀𝑡, 𝑙𝑛𝑋𝑡,
𝑙𝑛𝑇𝐵𝑡 for the import, export and trade balance equation respectively;
and𝑍𝑡 represents the exogenous variables from each equation.
5 see MacKinnon (2010), Critical Values for Cointegration Test.
6 The response surface estimates of critical values are provided in a table in appendix E.
Are B I Z E C O N S Q U A R T E R L Y
• 34 •
Equation (13) is used to analyse the short-run relationship between the dependent and
independent variables and the estimated coefficients give the short-run impact. The
coefficient of interest in this regression is 𝜑 which gives the speed of adjustment to a
long-run equilibrium relationship implied by cointegration regression when there is a
shock to this equilibrium in the short-run. A negative and statistically significant
𝜑 imply there is an adjustment to this long-run equilibrium relationship.
Although regressing a non-stationary variable on another non-stationary variable may
not be spurious because they are cointegrated, the standard errors from such
regressions are however potentially inefficient due to serial correlation and
endogeneity problem. Hence, we employ the Newey-west technique to correct the
OLS standard errors. Accordingly, the optimum lag length required for the Newey-
West regression is calculated using the truncation lag determination rule of thumb of
0.75T1/3
.
Also, the ECM model will be subjected to a series of diagnostic tests to confirm if
one or more of the following assumptions of: no serial correlation, homoscedasticity,
no autoregressive conditional heteroscedasticity (ARCH) effect and no omitted
variable is/are violated. The Breusch-Godfrey serial correlation LM test is used to
test for the presence of serial correlation in the error term. In the case where this
assumption is violated, the Newey-West method of correcting the OLS standard error
is employed in order to correct for the serial correlation. For the assumption of
homoscedasticity, we employ the Breusch-Pagan/Cook-Weisberg test for
heteroscedasticity. If the null hypothesis of a constant variance is rejected, we correct
for this problem by constructing a variance-covariance matrix that provides a
consistent estimator of the matrix. This adjustment will yield a heteroscedasticity
consistent standard errors that are known as White standard errors. A test that the
assumption of no ARCH effect is satisfied in our model is conducted using the LM
test for autoregressive conditional heteroscedasticity. The Ramsey's RESET test is
employed to determine if the assumption of no omitted variables is not violated in the
ECM model.
4. Empirical Results
Table 1 presents the results of the unit root tests conducted and the results show that
𝑙𝑛𝑀, 𝑙𝑛𝑋, 𝑙𝑛𝑌, 𝑙𝑛𝑌𝑊, 𝑙𝑛𝑅𝑃𝑋 and 𝑙𝑅𝑃𝑋 are non-stationary and integrated of order
one. The 𝑙𝑛𝑇𝐵 on the other hand is integrated of order zero I(0). However, given
the weak power of the test due to the short time period annual data available to us, for
the purpose of this study, 𝑙𝑛𝑇𝐵 was treated as an I(1) series.
Table 1. Unit Root test results for Aggregate Variables
Level First Difference
𝐼𝑚𝑝𝑜𝑟𝑡 -1.757 -6.348* I(1)
𝐸𝑥𝑝𝑜𝑟𝑡(𝐴) -1.848 -7.427* I(1)
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 35 •
𝐷 𝐼𝑛𝑐𝑜𝑚𝑒 1.151 -5.254* I(1)
𝑊 𝐼𝑛𝑐𝑜𝑚𝑒 -1.238 -4.224* I(1)
𝑅𝑃𝑀 -1.984 -6.218* I(1)
𝑅𝑃𝐸(𝐴) -0.940 -6.142* I(1)
𝑇𝑟𝑎𝑑𝑒 𝐵𝑎𝑙 -4.019* -6.858* I(0)
𝑅𝐸𝐸𝑅(𝐴) -2.487 -3.414** I(1)
NB: *, **, *** indicate statistical significance at 1%, 5%, and 10% level of significance respectively.
* (A) denotes Augmented Dickey-Fuller test.
*Critical values for model with constant are: -3.730 (1%), -2.992(5%) and -2,626(10%).
Having established that the variables are of the same order of integration, a
cointegration test was conducted, a cointegration analysis was done using the Engel-
Granger OLS-based cointegration technique. Equations (3) and (4) were estimated
using the OLS method and the estimated residuals from each equation were subjected
to a unit root test using equation (8). Also, to analyse the short-run dynamics and the
long-run dis-equilibrium, we estimated an error correction mechanism (ECM) model
to reconcile the short-run behaviour of import and export with their long-run
behaviour. The ECM models were subjected to rigorous diagnostic test to determine
if they are well specified and fit for the purpose of this study. The results of the tests
are reported in (Table 2).
Table 2. Cointegration and ECM Result for the Import and Export Function
Variables Import Function Export Function
Long-Run
Const.
-13.5783
(7.2649)
-21.7770
(15.7059)
Income 0.6026*
(0.2324)
0.8451***
(0.5011)
Relative price -0.6456*
(0.2406)
-0.0979
(0.1183)
Dickey-Fuller test -5.60* -7.54*
R2 0.7732 0.3587
Short-run
Const.
0.0145
(0.0495)
0.0119
(0.0526)
Income -0.3661
(0.7557)
-0.4356
(1.6850)
Are B I Z E C O N S Q U A R T E R L Y
• 36 •
Relative price -0.4215*
(0.1508)
-0.1767*
(0.0729)
ECM(t-1) -0.9254*
(0.1875)
-0.8323*
(0.1101)
F-stat 14.14 20.30
Diagnostic test
Heteroscedasticity 𝑥21] 5.27 0.26
Serial correlation
𝑥2[2]
3.23 5.246
ARCH TEST 𝑥2[2] 0.028 1.836
OV test F [3,21] 2.19 0.67
NB: *, **, *** denotes statistical significance at 1%, 5%, and 10% level of significance respectively.
-The t critical value with 26 degree of freedom for a two tail test are: 1%(±2.479) 5%(± 2.056) and
10%(±1.706) respectively.
-The standard errors reported are corrected for serial correlation using Newey-West regression.
-The optimum lag length used in the Newey-West regression is approximately 2, calculated using the
truncation lag determination rule of thumb of 0.75T1/3.
-Dickey Fuller test is the t(tau) value for the unit root test conducted for each residual.
-Income in the import function is the domestic income while income in the export function is the world
income.
-The MacKinnon critical value for cointegration test for three I(1) series with no trend are: -4.829(1%),
-4.047(5%), and -3.670(10%).
The Dickey-Fuller test in each table is the t(tau) value from the unit root test
conducted for each set of residuals, while the estimated coefficients are the long-run
elasticities. The OLS standard errors are not reported because the series used are non-
stationary which implies that the regression suffers from serial correlation and
endogeneity problems. Hence, we employed the Newey-west technique to correct for
these problems.
The estimated t(=tau) value, from the unit-root test for the aggregate import demand
and export supply equation residuals are -5.60 and -7.54 respectively, which are both
significant at 1% level using the MacKinnon critical values for cointegration. These
stationary residuals imply the existence of cointegration, which suggest a long-run
equilibrium relationship amongst the import and export function variables.
It is also of interest to determine how the volume of imports and exports respond to
changes in relative prices and income in the long-run which are provided by the long-
run elasticities. In terms of the aggregate relative price elasticities, both signs are
consistent with economic theory. However, only the price elasticities of demand for
imports is statistically significant in the long-run.
From our estimation, if the Naira falls by 1%, it is expected that Nigerians will
reduce their demand for foreign goods by 0.65% which is significant at all
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 37 •
conventional levels. This decline in demand for foreign goods, everything else
constant, is more likely to result into inflation than having a positive impact on the
economy. This is because Nigeria import most of the raw materials used for domestic
production and this increase in the cost of importation will increase the cost of
domestic production which will results into inflation in the long-run.
More significant in our analysis in the long-run are the income elasticities. The
domestic and world income suggest an inelastic and significant relation both on the
volume of imports and exports in the long-run. Such that in reaction to a similar 1%
increase in domestic income, Nigerian imports from the rest of the world will be
increased by 0.60% while exports to the world will also increase by 0.85%. The value
of the adjusted R2 for the import and export functions of 0.77 and 0.36 respectively,
indicates that variation in relative prices and incomes explain about 77% of the
variation in imports and 36% of the variations in exports. The economic implication
here is that, on average, in the long-run, income has a more significant and
dominating effects on the trade than prices. Also, the import functions are more fit
for the purpose of this study than the export function.
The short-run elasticities estimated and the results of the regression diagnostic tests
are reported in (Table 3). The diagnostic tests result reveals that the import demand
specification is heteroscedastic, which implies that the standard errors from this OLS
regression are inefficient. This was corrected using the robust option to obtain
heteroscedastic consistent standard errors reported in the table.
Unlike the long-run analysis, only price has a significant impact on imports and
exports in the short-run. In response to a 1% fall of the Naira, Nigeria’s import is
expected to decrease by 0.42% while Nigeria’s export supply to the world is expected
to increase by 0.17% which are both statistically significant at 5%. The economic
implication from our estimates is that price matters in the short-run but not in the
long-run for exports.
The error correction mechanism coefficient of the import and export functions,
though negative and significant at the 5%, indicate the rapid adjustment to shocks in
the short-run which is implausible in international trade. The coefficient suggests
that when the volume of imports or exports deviates from their equilibrium value
given by the cointegration regression, all of this deviation will be corrected for within
a year. These instantaneous adjustments to shocks within one year appears
implausibly fast and assumes no adjustment is required to attain long-run equilibrium
indicating that this model may not reflect the reality of Nigeria’s trading
relationships. Therefore, our estimates will be interpreted with utmost cautions to
avoid misleading inferences.
Having laid the background for our analysis, we proceed to testing the M-L condition
using the relative price elasticities. In acknowledgement of the insignificance of the
relative price elasticity in the long-run, we test for the M-L condition based on the
existence of a long-run equilibrium relationship amongst the variables. Also, bilateral
M-L condition tests are conducted while admitting that most of the price elasticities
Are B I Z E C O N S Q U A R T E R L Y
• 38 •
are insignificant in the long-run with reverse signs which might give unreliable
results. The results for the M-L condition tests are reported in (Table 3).
The result from testing the null hypothesis (that the sum of the price elasticities is
equals to one in absolute terms) shows that we cannot reject the null at a 5% level of
significance. The calculated t-value and prob value for rejecting the null are -0.957
and 0.1685. Hence, we tentatively conclude that the condition is not satisfied for
Nigeria in this study. This result corroborates findings from Loto (2011). However,
our lack of support for the M-L condition could be because the estimated export price
elasticities are insignificant in the long-run. Overall, we tentatively reject the M-L
condition for Nigeria due to the various limitations from our estimation and models.
Table 3. Test of Marshall-Lerner Condition
Trading Partners t-values
(prob)
Decision Rule
World -0.957
(0.1685)
Not satisfied
NB: *, **, *** denotes statistical significance at 1% , 5%, and 10% level of significance respectively,
and -The prob value for rejecting the null are reported in the brackets.
The trade balance analysis was done to examine the long-run and the short-run
relationship between the trade balance and the real exchange rate using the Engle-
Granger OLS-based cointegration technique and error correction mechanism (ECM)
modelling technique. The results reported in table 9 confirm the existence of a long-
run equilibrium relationship between the real effective exchange, domestic income,
foreign income and the trade balance in Nigeria. However, the elasticities indicate
that real effective exchange rate do not have significant impact on the trade balance
in the long-run. This provides support to our findings from investigating the M-L
condition that devaluation will not improve the trade balance in Nigeria in the long-
run.
In addition, the lower panel of table 4 presents the results of the regression diagnostic
tests and the ECM regression which captures the short-run dynamics. The regression
diagnostic test results are satisfactory, but the insignificant short-run real exchange
rate elasticities imply that devaluing the Naira will have no impact on the trade
balance in the short-run. The error correction model coefficient of -0.85 is
statistically significant at 5% level of significance. This is similar to the estimated
speed of adjustment for the aggregate import and export function. This suggests that
this trade balance ECM model suffers from aggregation bias and not relying on these
estimates might led to misleading conclusions.
However, our finding from this empirical analysis suggests that devaluation cannot
improve the trade balance both in the long-run and also in the short-run for Nigeria.
This is consistent with the finding of Ogundipe, Ojeaga, and Ogundipe (2013), Eke,
Eke and Obafemi (2015), and Olaniyi (2019).
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 39 •
Table 4. Cointegration and ECM Regression Results for the Trade Balance Analysis
Variables Cointegration Regression ECM Regression
Const. -14.4645
(10.1927)
4.3994
(6.3147)
Income -0.1762
(0.2454)
1.3143
(0.7633)
Relative price -0.1459
(0.1155)
-0.2165
(0.1189)
World Income 0.6769
(0.5346)
-0.1422
(0.2018)
D. F. Test -4.12*
ECM(t-1) -0.8525*
(0.2012)
R2 0.1891
F-stat 5.86
Diagnostic test
Heteroscedasticity 𝑥2[1] 0.09
Serial correlation 𝑥2[2] 1.643
ARCH TEST 𝑥2[2] 2.890
OV test F [3,21] 0.60
NB:-The cointegration standard errors reported are corrected for serial correlation using Newey-West
regression.
-The optimum lag length used in the Newey-West regression is 2, calculated using the truncation lag
determination rule of thumb of 0.75T1/3.
-Dickey Fuller test is the t(tau) value for the unit root test conducted for each residual.
-The MacKinnon critical value for cointegration test for four I(1) series with no trend are: -5.312(1%), -
4.497(5%), and -4.104(10%).
-A decrease of the real exchange rate implies a devaluation of the Naira.
-The reported ECM standard errors are corrected for violation of the critical assumptions when
necessary.
5. Conclusion and Policy Implication
The notion that a devaluation will improve the trade balance in the long run has for
decades been a fundamental tenet and has formed the basis of several empirical
research. Several studies have analysed and assessed the validity of this tenet using
data from different countries and using different econometric techniques with mixed
and conflicting results. This is also the case in Nigeria where there is no consensus as
to what the impact of a devaluation will have on the trade balance in Nigeria in the
Are B I Z E C O N S Q U A R T E R L Y
• 40 •
long-run despite several decades of research. Hence the need for a re-examination
given the Nigeria’s current economic situation.
This paper re-examined this contentious issue by investigating the Marshall-Lerner
(M-L) condition for Nigeria. It also examined the long and the short-run relation
between the trade balance in Nigeria and the real exchange rate. The results from
testing for the M-L condition failed to find support for the condition in Nigeria.
Hence, it was tentatively concluded that the M-L condition does not appear to hold
persuasively for Nigeria. In addition, the empirical analysis of the relationship that
exists between the trade balance and the real exchange rate suggests that either in
long-run or in the short-run, the real effective exchange rate does not appear to have
any significant impact on Nigeria’s trade balance.
Given these results, it was concluded that the M-L condition does not hold for
Nigeria and a real devaluation of the Naira would have no significant improvement
on the trade balance both in the short-run and in the long-run. In addition, the
findings from this research corroborate an important economic principle that income
effects dominate price effects. This is evident from the long-run analysis which
revealed that even when prices are flexible, income elasticities are larger and have
more significant effects on trade.
A suggestion for further research, based on the findings and limitations encountered
in this research, is that that further investigation of the M-L condition for Nigeria
needs to be conducted on bilateral basis. This allows for the heterogeneity in bilateral
trading transaction to be captured which will be very informative for trade policy
purpose. Also, more can be done on this literature for Nigeria by conducting research
that analyses oil exports separately from non-oil exports, covering longer period
beyond 1986, to adequately account for the role of oil exports on the trade balance in
Nigeria. In addition, a policy recommendation from the findings is that monetary
authorities should employ other policy instrument to improve the trade balance as the
Nigeria trade structure does not support a devaluation improving the trade balance of
the nation.
References
Alexander, S.S. (1952) ‘Effects of a Devaluation on a Trade Balance’, International
Monetary Fund, 2 (2), pp. 263-278, JSTOR [Online]. Available at:
http://www.jstor.org/stable/3866218 (Accessed: 6 Jun 2016).
Anoke, C.I., Odo, S.I, & Ogbonna, B.C. (2016) ‘Effect of Exchange Rate
Depreciation on Trade Balance in Nigeria’, Journal of Humanities and Social
Science, 21(3), pp. 72-81. [Online]. Available at: www.iosrjournals.org
(Accessed: 17 Jun 2016).
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 41 •
Appleyard, D.R. & Field, A.J. (2001) ‘Price Adjustment and Balance-of-Payments
Disequilibrum’4th edn., International Economics. Boston, Mass.: Boston,
Mass. : McGraw-Hill/Irwin, pp. 527-551.
Arize, C.A. (1994) ‘Cointegration Test of a Long-run Relation Between the Real
Effective Exchange Rate and the Trade Balance’, International Economic
Journal, 8(3), pp. 1-9, [Online] DOI: 10.1080/10168739400000001
(Accessed: 4 Apr 2016).
Bahmani-Oskooee, M. (1991) ‘Is there a Long-run Relation Between the Trade
Balance and the real Effective Exchange Rate of LDCs?’, Economic Letter,
36, Elsevier Science Publishers, pp.403-407.
Bahmani-Oskooee, M. & Alse, J. (1994) ‘Short-run versus Long-run Effects of
Devaluation: Error-Correction Modelling and Cointegration’, Eastern
Economic Journal, 20(4), pp. 453-464, JSTOR [Online]. Available at:
http://www.jstor.org/stable/40325598 (Accessed: 22 Apr 2016).
Bahmani-Oskooee, M. (1998) ‘Cointegration Approach to Estimate the Long-run
Trade Elasticities in LDCs’, International Economics Journal, 12(3), pp. 89-
96, [Online] DOI: http://dx.doi.org/10.1080/10168739800000031 (Accessed:
21 Mar 2016).
Bahmani M., Harvey, H. & Hegerty, S.W. (2013) ‘Empirical Tests of the Marshall-
Lerner Condition: A literature Review’, Journal of Economics Studies, 40(3),
pp. 411-433, [Online] DOI:
http://dx.doi.org/10.1108/01443581311283989 (Accessed: 21 Mar
2016).
Boyd, D., Caporale, G.M. & Smith, R. (2001) ‘Real Exchange Rate Effects on the
Balance of Trade: Cointegration and the Marshall-Lerner Condition’
International Journal of Finance and Economics, 6(3), pp. 187-200.
Central Bank of Nigeria Statistical Bulletin, (2016), External Sector Statistics,
Available at: https://www.cbn.gov.ng/documents/Statbulletin.asp
Dornbusch, R. (1975) ‘Exchange Rate and Fiscal Policy in a Popular Model of
International Trade’, The American Economic Review, 65(5), pp. 859-871,
JSTOR [Online]. Available at: http://www.jstor.org/stable/1806625
(Accessed: 15 Aug 2016).
Eita, J.H. (2013) ‘Estimation of the Marshall-Lerner Condition for Namibia’,
International Business and Economics Research Journal, 12(5), pp. 511-518.
Eke, I. C., Eke, F.A. & Obafemi, F.N. (2015) ‘Exchange Rate Behaviour and Trade
Balances in Nigeria: An Empirical Investigation’, International Journal of
Humanities and Social Science, 5(8(1)), pp. 71-78.
Enders, W. (1995) Applied econometric time series. New York: New York: Wiley.
Are B I Z E C O N S Q U A R T E R L Y
• 42 •
Engel, R.F. & Granger, C. W. (1987) ‘Cointegration and Error Correction:
Representation, Estimation, and Testing’, Econometrica, 55(2), pp.251-276.
Evans, O. (2019). The Curious Case of Petro-Monetary Transmission Mechanism in
Oil-Producing Countries: An Analysis of the Effect of Oil Price on Inflation
in Nigeria. Iranian Economic Review. DOI: 10.22059/IER.2019.70884.
Evans, O., Adeniji, S., Nwaogwugwu, I., Kelikume, I., Dakare, O., & Oke, O. (2018).
The relative effect of monetary and fiscal policy on economic development
in Africa: a GMM approach to the St. Louis equation. BizEcons Quarterly, 2,
3-23.
Eze, T.C. & Okpala, C.S. (2014) ‘Quantitative Analysis of the Impact of Exchange
Rate Policies on Nigeria’s Economic Growth: A Test of Stability of
Parameter Estimates’, International Journal of Humanities and Social
Science, 4(7), Centre for Promoting Ideas, USA, pp. 265-272.
Goldstein, M. & Khan, M.S. (1978) ‘The Supply and Demand for Exports:
Simultaneous Approach’, The Review of economics and Statistics, 60(2), pp.
275-286, JSTOR [Online]. Available at: http://www.jstor.org/stable/1924981
(Accessed: 20 Apr 2016).
Gujarati, D. (2011) Econometric by Example, 1st edn. Palgrave Macmillan.
Hacche, G. & Townend, J.C. (1981) ‘Monetary Models of Exchange Rates and
Exchange Market Pressure: Some General Limitations and an Application to
Sterling's Effective Rate’, Reviewof world Economics, 117(4), pp. 622-637,
[Online] DOI:10.1007/BF02708114 (Accessed: 13 Aug 2016).
Himarios, D. (1989) ‘DO Devaluation Improve the Trade Balance? The Evidence
Revisited’, Economic Inquiry, Vol.XXVII, Periodicals Archive Online, pp.
143-168.
Hirschman, A.O. (1949) ‘Devaluation and the Trade Balance: A Note’, The Review
of Economics and Statistics, 31(1), pp.50-53, JSTOR [Online]. Available at:
http://www.jstor.org/sici?sici (Accessed: 16 Jun 2016).
Igue, N.N. & Ogunleye, T.S. (2014) ‘Impact of Real Exchange Rate on Trade
Balance in Nigeria’, African Development Review, 26(2), Blackwell
Publishing Ltd. pp. 347-358.
International Monetary Fund, (2016) ‘IMF DATA’, Available at:
http://www.imf.org/en/Data.
Lerner, A.P. (1944) ‘Economics of Control: Principles of Welfare Economics’. The
Macmillan Company. N.Y.
Loto, M.A. (2011) ‘Does Devaluation Improve the Trade Balance of Nigeria? (A
Test of the Marshall-Lerner Condition)’, Journal of Economics and
International Finance, 3(11), pp. 624-633, [Online]. Available at:
http://www.academicjournals.org/JEIF (Accessed: 04 Apr 2016).
Currency Devaluation and Trade Balance Nexus: A Test of Marshall-Lerner Condition in Nigeria
• 43 •
Marshall, A. (1923) ‘Money, credit and commerce’, London, Macmillan.
Miles, M.A. (1979) ‘The Effects of Devaluation on the Trade Balance and the
Balance of Payments: Some New Results’, Journal of Political Economy,
87(3), pp. 600-620. JSTOR [Online]. Available at:
http://www.jstor.org/stable/1832025 (Accessed: 16 Jun 2016).
Odili, O. (2014) ‘Exchange Rate and Balance of Payment: An Autoregressive
Distributed Lag (ARDL) Econometric Investigation on Nigeria’, Journal of
Economics and Finance, 4(6), pp. 21-30, ISOR [Online]. Available at:
http://www.iosrjournals.org (Accessed: 7 Apr 2016).
Ogbonna, B.C. (2011) ‘The Impact of Exchange Rate Variations on Trade Balance:
Evidence from Nigeria, 1970-2005’, JORIND, 9(2), pp. 393-403, [Online].
Available at: www.ajol.info/journals/jorind (Accessed: 7 Apr 2016).
Ogundipe, A.A. Ojeaga, P. & Ogundipe, O.M. (2013) ‘Estimating the Long-Run
Effects of Exchange Rate Devaluation on the Trade Balance of Nigeria’,
European Scientific Journal, 9(25), pp. 233-249.
Oladipupo, A.O. & Onotaniyohuwo, F.O. (2011)’ Impact of Exchange Rate on
Balance of Payment in Nigeria’, African Research Review, 5(4), pp. 73-88,
Indexed African Journals [Online]. Available at: www.ajol.info(Accessed:
07 Apr 2016).
Olaniyi, E. (2019). The effects of US-China trade war and Trumponomics. Forum
Scientiae Oeconomia, 7(1), 47-55.
Sinha, D. (2001) ‘A note on trade elasticities in Asian countries’, The International
Trade Journal, 15(2), pp. 221-237, [Online] DOI:
http://dx.doi.org/10.1080/088539001316901042 (Accessed: 4 Apr 2016).
United Nations Conference on Trade and Development (UNCTAD) Statistics,
International Trade in goods and Services, Available at:
http://unctadstat.unctad.org/wds/ReportFolders/reportFolders.aspx?IF_
ActivePath=P,15912&sCS_ChosenLang=en.
World Bank (2016) ‘World Development Indicators’, World Bank, [Online].
Available at: http://data.worldbank.org (Accessed: 04 Jun 2016).
This work is licensed under a Creative Commons Attribution-NonCommercial-
ShareAlike 3.0 Unported license. To view a copy of this license, visit
http://creativecommons.org/licenses/by-nc-sa/3.0